fics games collection alternative

[Dann Corbit]

You can collect all FICS games since 1999 as bzip2 compressed PGN here:
http://marcelk.net/logics/
Majority are collected by torrent.

Or you can collect them as bzip2 compressed SCID games here:
http://cap.connx.com/scid/

For 2008 and 2009, I had to divide the game collections into 2 parts (2008a, 2008b, 2009a, 2009b), because a SCID database could not contain the entire set of games for the complete year.

For 2010 because we are not yet at the half-way point in the year, I included the games as month-by-month.

I also added this:
fics_elo2000.sg3.bz2
fics_elo2000.si3.bz2
fics_elo2000.sn3.bz2

which consists of all FICS games where both players have Elo values greater than or equal to 2000 (most FICS games are between low rated opponents).

Warning:
These game collections are titanic in size, and generally of very low quality. I do see them as useful (for instance) in learning what kind of mistakes low-end chess players tend to make. They could also be used for discovery of trends of ability by rating, since these collections have only rated games (no guest games in the set).

[Volker Pittlik]

...
Warning:
These game collections are titanic in size, and generally of very low quality. I do see them as useful (for instance) in learning what kind of mistakes low-end chess players tend to make. They could also be used for discovery of trends of ability by rating, since these collections have only rated games (no guest games in the set).

Have you tried to make a book out of the games? According to my own tests the resulting books are deep but of a lousy quality (engines get significant lower ratings). I tested it at fast time controls what would mean that saving time does not compensate the bad quality of moves.


Volker

[beachknight]

You can collect all FICS games since 1999 as bzip2 compressed PGN here:
http://marcelk.net/logics/
Majority are collected by torrent.

Or you can collect them as bzip2 compressed SCID games here:
http://cap.connx.com/scid/

For 2008 and 2009, I had to divide the game collections into 2 parts (2008a, 2008b, 2009a, 2009b), because a SCID database could not contain the entire set of games for the complete year.

For 2010 because we are not yet at the half-way point in the year, I included the games as month-by-month.

I also added this:
fics_elo2000.sg3.bz2
fics_elo2000.si3.bz2
fics_elo2000.sn3.bz2

which consists of all FICS games where both players have Elo values greater than or equal to 2000 (most FICS games are between low rated opponents).

Warning:
These game collections are titanic in size, and generally of very low quality. I do see them as useful (for instance) in learning what kind of mistakes low-end chess players tend to make. They could also be used for discovery of trends of ability by rating, since these collections have only rated games (no guest games in the set).

Dear Dann,

Thank you. Unfortunately I have a limited monthly download quota and
therefore I could not download and study all of them now.

I want to add another reason for downloading FICS (and ICC,
playchess, other chess servers like net-chess) games:

Almost all commentators gave 14. ... Rg8 as novelty in the third WCh
game. In my research database, I have found a FICS game and
playchess games that suggests 15 h4 is the real novelty. FICS game:

[15.0-0 Bc5 16.Bxc5 Nxc5 17.b4 Nd7 18.f4 Ke7 19.f5 Bh5
20.h3 Rgc8 21.Ba2 exf5 22.exf5 f6 23.Nd5+ Kf8 24.Nc7 Rab8
25.Ne6+ Ke7 26.Nxg7 Bf7 27.Be6 Bxe6 28.fxe6 Ne5 29.g4 b5
30.Rxc8 Rxc8 31.axb5 axb5 32.Ra1 Rc7 33.Ra6 Nd3 34.Nf5+ Kf8
35.Ra8+ 1-0 Dampfwalze - darkiest/ FICS 2008]

Best,

[Dann Corbit]

...
Warning:
These game collections are titanic in size, and generally of very low quality. I do see them as useful (for instance) in learning what kind of mistakes low-end chess players tend to make. They could also be used for discovery of trends of ability by rating, since these collections have only rated games (no guest games in the set).

Have you tried to make a book out of the games? According to my own tests the resulting books are deep but of a lousy quality (engines get significant lower ratings). I tested it at fast time controls what would mean that saving time does not compensate the bad quality of moves.


Volker

I would never use the unfiltered games for books.
Imagine the chess games of one million randomly selected high school kids. That's what these games would look like.